[gmx-users] CHARMM and FEP calculations

[Mark Abraham]

On 24/08/2012 11:28 PM, Wright, Louise wrote:
> Hi
> I asked a similar question a year ago about implementing CMAP in FEP calculations for gromacs 4.5.4.  At the time I could not get the simulation to run with CMAP as it came up with:
> Fatal error:
>> Function type CMAP Dih. not implemented in ip_pert
> Helpfully, Mark suggested a modification to the code to get round this issue, but I am unclear as to whether in the newer version of gromacs (4.5.5) CMAP terms can actually be scaled from state A to state B or whether for all values of lambda this part of the potential remains in state A even if I were to manually write state B CMAP terms in the topology file and the code no longer gives a warning.

I expect the latter, since 
http://www.gromacs.org/About_Gromacs/Release_Notes/Versions_4.5.x makes 
no mention of CMAP and FEP. Someone would have to want to do the coding 
and testing work to make this possible.

> In addition, for CHARMM, and other FFs, there are some dihedrals which have multiple potentials, using the dihedral function type 9.  I have manually specified the state B parameters for these (and all other dihedrals) in my topology file but get the error message:
>
>    Cannot automatically perturb a torsion with multiple terms to different
>    form.
>    Please specify perturbed parameters manually for this torsion in your
>    topology!

This error pertains to what grompp can look up in ffbonded.itp for this 
interaction, and is perhaps over-zealous, but without seeing the 
relevant file fragments it's hard to say anything.

Mark

>
> Is there a way to implement scaling of these multiply defined dihedrals in FEP calculations?
>
> Many thanks for your help
>
> Louise
>